Quantum Gold Premium
- Uzmanlar
- Sürüm: 6.1
- Güncellendi: 1 Kasım 2026
- Etkinleştirmeler: 5
QuantumGoldPremium AI Train 6.1
An XAUUSD trading system for MetaTrader 5 with 56 strategy branches and 9 embedded ONNX models.
Two gold candles can close at the same price and tell different stories. One may extend a move supported by volume and higher-timeframe momentum. The other may leave a long lower wick as buyers absorb selling pressure. QuantumGoldPremium uses those differences to decide whether a setup qualifies for entry.
Version 6.1 combines 54 technical strategy branches with dedicated BUY and SELL Momentum strategies. Nine models trained on historical gold data evaluate selected setups directly inside MT5. Each model has a defined role, and each approved strategy uses its own decision threshold.
| Specification | Details |
|---|---|
| Platform | MetaTrader 5 |
| Instrument | XAUUSD, including compatible broker symbols such as XAUUSDm |
| Chart timeframe | M5, with additional context from M15, H1, and H4 |
| Strategy engine | 54 technical branches plus BUY Momentum and SELL Momentum |
| AI engine | 9 embedded ONNX models running locally on CPU |
| Execution | One EA-managed position per symbol at a time, with an initial Stop Loss and strategy-defined Take Profit |
| Default configuration | AI filtering enabled, equity-based sizing selected, research logging off |
| Installation | One EX5 file. All models and strategy settings are included. |
How the nine AI models are used
Seven strategy-specific models evaluate SELL_2, SELL_4, SELL_9, SELL_11, SELL_21, BUY_RSID, and BUY_8.1. When one of these strategies produces an eligible setup, its model scores the market conditions visible before entry. The EA compares that score with the threshold researched for that particular strategy. Other branches follow their own technical entry rules.
Two specialist models focus on SELL Momentum. A support vector machine with an RBF kernel evaluates the original momentum setup. A separate logistic regression model evaluates additional entry opportunities under its own threshold. This separation lets the EA assess an established signal and an expanded setup against their respective historical examples. BUY Momentum follows its dedicated technical rules.
The training process starts with what the EA knew at the time. Pre-entry feature snapshots are matched with subsequent executions and completed trades for offline training. Models are then evaluated on later historical periods. Inside MT5, feature-schema checks and reference predictions verify that each embedded ONNX model receives the expected inputs and reproduces its reference scores. Date-based activation controls when each model becomes eligible during a historical replay.
What the EA reads before entry
- Trend and momentum across timeframes. MACD, moving-average slopes, and the position of price relative to those averages help put an M5 signal in the context of M15, H1, and H4.
- Volatility and price location. ATR-normalized measurements express candle size, recent movement, spread, and distance from technical levels relative to current volatility. TMA and band relationships add context about where the setup formed.
- Candle structure and participation. Bodies, upper and lower wicks, closing location, and relative volume help distinguish directional movement from rejection. Momentum strategies also use ADX and WaveTrend features.
- Agreement between strategies. If SELL_9 rejects a setup and SELL Momentum proposes a closely matching trade on the same tick, the EA checks the entry, SL, and TP together. A second strategy label alone is insufficient to qualify the same trade.
One example is SELL_1. When a completed M5 candle leaves a pronounced lower wick while H1 momentum remains positive, the entry filter can decline the short. SELL_9 examines a different condition: price has traveled back and forth over 20 M5 candles, made little net progress, and M15 momentum offers limited support for a sell. Both decisions take place before the order is sent.
Position sizing and trade management
Equity-based risk is the default, with RiskPercent set to 5.0%. The EA estimates the loss from entry to the planned Stop Loss using the broker’s contract specifications, then calculates volume from current equity and the selected percentage. Volume is rounded down to the permitted lot step. If the calculated size is below the broker’s minimum, the EA can use the minimum lot, so planned risk can exceed the selected percentage. Slippage, gaps, and trading costs can also affect the realized result.
Base-lot sizing uses the LOT_SIZE you select, with a default of 0.1 lot. When the Stop Loss is more than 8 gold-price units from entry, the EA reduces volume according to its distance-based sizing rule.
Open positions are managed through the strategy’s SL and TP rules. EnableTrailingSl selects continuous trailing or the original stepped SL adjustments. News-avoidance intervals and low-volume session settings also influence when the EA can open a position.
AI settings
| Mode | How it operates |
|---|---|
| FILTER — default | Applies approved model thresholds and the additional SELL entry checks before qualifying an order. |
| SHADOW | Calculates eligible model scores while the technical strategy controls entry. |
| COLLECT | Supports structured signal and trade collection for offline research when data logging is enabled. |
| OFF | Runs the technical strategy engine with its built-in technical filters. |
AI filtering works with research logging turned off. Enable EnableAIDataLogging only when you want to save research files. Live operation uses the models embedded in the release. Model training takes place offline.
Getting started
Attach QuantumGoldPremium AI Train to your broker’s XAUUSD M5 chart and select Inputs → Reset to load the release defaults. Review LotSizingMode and RiskPercent , then enable Algo Trading when your configuration is ready. The Premium edition keeps individual strategy switches hidden and exposes the settings needed for sizing, trade management, scheduling, and AI operation.
One EX5 file is all you need. No external ONNX files, Python installation, or preset is required. For the installation manual and setup assistance after purchase, contact me through MQL5 private messaging.
